Page 252 - Fundamentals of Gas Shale Reservoirs

232 PASSIVE SEISMIC METHODS FOR UNCONVENTIONAL RESOURCE DEVELOPMENT

trace processing stage. Coherent noise remaining in the trace‐ discontinuity and that the cloud of high activity is approxi­
processed signal is usually very apparent if not removed, mately centered on the main discontinuity.
indicating the need for more refined trace processing. Fracture imaging uses the above empirical observations
The key to this method is that a very large number of on the relationship between seismic energy emission and the
image volumes are computed and then summed. A voxel that damage zone to operate on the stacked semblance volumes
emits seismic energy not periodically but continuously for a to extract surfaces representing the fracture/fault networks.
long period of time will have a very large cumulative activity The workflow for computing the emission volumes and
value. The total seismic energy emitted from each voxel is accumulating them into a final volume for input to the frac­
accumulated over tens to hundreds of thousands of time ture network computation is as follows:
windows. The final volume will show the cumulative emis­
sions in each voxel. Because statistical measures such as 1. A threshold is applied to a stacked semblance volume
semblance or covariance are typically used as an activity to remove the low‐level background. Although the
measure the cumulative activity values are relative—they are background may contain low‐level signal, it also
not an absolute measure of energy. Note that random noise is contains false structure resulting from noise. The
not completely removed because random processes are not result of this step is sinuous clouds of high cumulative
uniform—they exhibit clustering and anticlustering. Some activity. These images can be used to provide an
false structure is imaged at a very low activity level. Practical alternative measure of the SRV (Lacazette et al.,
experience and modeling studies show that spatially stable 2014; Figure 10.21).
signal overwhelms false structure imaged by noise over 2. In the final step, the central surfaces of the high‐activity
typical stacking time intervals. clouds are computed. These central surfaces are repre­
Different strategies may be used to combine the volumes. sented as either raster images, single‐voxel thick
The simplest is to simply sum all of the volumes over the surfaces, or as vector images. The vector images are
time period of interest. Alternatively, the volumes may be tessellated surfaces. A tessellated surface is a contin­
edited statistically and then substacked into fewer volumes uous surface comprising triangles that share vertices
depending on product desired (Lacazette et al., 2014). The and edges. The tessellated surfaces can be derived either
result is a single depth volume for the entire time period of from the raster images or directly from the stacked sem­
interest, such as a frac stage, that shows the spatial distribu­ blance volumes (Lacazette et al., 2014 Copeland and
tion and relative intensity of activity. Lacazette, 2015)). Figure 10.22 shows the most ener­
getic region of a fracture image that represents the likely
10.5.5 Direct Imaging of Fracture Networks extent of proppant distribution. Tessellated surfaces are
commonly stored as ASCII data in the nonproprietary
Fracture imaging methods can directly image induced and TSURF format, which is read and written by a wide
natural fracture flow paths in oil and gas reservoirs (Geiser variety of software packages.
et al., 2006; Geiser et al., 2012; Lacazette and Geiser, 2013;
Lacazette et al., 2013; Lacazette et al., 2014; Sicking et al., More detailed description and discussion of nuances of the
2014). This relatively new method computes fracture images method are provided by Geiser et al. (2012), Sicking et al.
from cumulative seismic activity volumes, which are described (2012), Lacazette et al. (2013), and Lacazette et al. (2014).
in the previous section. Consequently, fracture imaging is yet
another extension of the SET workflow. Fracture imaging 10.5.6 Comparison of Downhole Hypocenters
benefits from previously described enhanced sensitivity of and Fracture Images
cumulative activity imaging.
The method relies on the known behavior of failure This section describes an example of hypocenters derived
processes in the Earth’s brittle crust. Rock mechanics from conventional downhole monitoring and fracture maps
theory, field studies (e.g., Vermilye and Scholz, 1998), and determined by the methods described in Section 10.5.4.
experiments (e.g., Janssen et al., 2001) show that large Figure 10.23 shows the near well zone for one frac stage.
fractures are embedded in clouds of smaller fractures and The downhole hypocenter detections are overlain on the
seismic emissions that become exponentially more fracture image mapped from the surface microseismic. There
intense proximal to the main fracture surface. The three is excellent agreement between the fracture image and the
parameters of the fracture/fault systems that control the downhole hypocenters. The azimuth of the fractures agrees
emissions are the distribution of the size and frequency of with the SHmax and Shmin for this area. Figure 10.24 shows
fractures, the fracture size distribution within the fracture/ the fault tracks extracted from the surface reflection seismic
fault zone, and the fracture/fault kinematics. The model for data with overlays of the fracture network from surface
fracture imaging predicts acoustic energy produced by the microseismic for the entire well and all of the downhole
fracturing reaches a maximum in the vicinity of the main hypocenters for the well. The large area fracture image maps